The current landscape presents the most serious test to date for open source AI. Nathan Lambert argues that the window for open models to remain competitive is closing. This pressure forces a reckoning over whether open weights can keep pace with proprietary scaling. Practitioners must now evaluate if open alternatives remain viable.